«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

期数:

2025年02期

页码:

24-32

栏目:

媒体计算

出版日期:

2025-02-10

文章信息/Info

Title:

An Improved DETR Model for Forest Fire Smoke Recognition

文章编号:

1673-629X(2025)02-0024-09

作者:

颜谨; 肖满生; 王瑶瑶; 朱泽宇

湖南工业大学 计算机学院,湖南 株洲 412000

Author(s):

YAN Jin; XIAO Man-sheng; WANG Yao-yao; ZHU Ze-yu

School of Computer Science and Engineering,Hunan University of Technology,Zhuzhou 412000,China

关键词:

烟雾检测; DETR; 多尺度特征信息; 注意力机制; CutMix

Keywords:

smoke detection; Detection Transformer; multi-scale feature information; attention mechanism; CutMix

分类号:

TP391

DOI:

10.20165/j.cnki.ISSN1673-629X.2024.0296

摘要:

针对传统的基于卷积神经网络(CNN)的森林火灾烟雾检测中,需要大量人工设计构件、对复杂森林场景中不明显的小烟雾检测能力较差等问题,提出了一种改进 DETR(Detection Transformer)的森林火灾发生早期的烟雾检测模型。 首先,使用 DETR 作为基线,将多尺度对比局部特征模块(MCCL)和密集金字塔池化模块(DPPM)集成到特征提取模块中,用于感知小烟雾或不明显烟雾的特征;然后,提出了一种边界框迭代组合方法,生成能够覆盖整个烟雾对象的精确包围盒,以提高检查精度、减少误检与漏检;最后,利用 CutMix 数据增强扩充森林火灾烟雾数据集,对该方法进行评价。 理论分析与实验表明,改进后的模型对森林火灾烟雾的检测精度明显高于主流模型,与传统 DETR 模型相比,该模型的 mAP(平均精度均值)提高了 4. 4% ,AP50精度提高了 3. 8% 。

Abstract:

Aiming at the problems of the traditional convolutional neural network (CNN) based forest fire smoke detection,such as the need for a large number of artificial design components and poor ability to detect small smoke which is not obvious in complex forest scenes,an improved DETR ( Detection Transformer) smoke detection model in the early stage of forest fires is proposed. Firstly,the multi-scale contrast local feature module (MCCL) and dense pyramid pool module (DPPM) are integrated into the feature extraction module using DETR as the baseline for sensing the features of small or invisible smoke. Then,a boundary box iterative combination method is proposed to generate an accurate bounding box that can cover the entire smoke object,so as to improve inspection accuracy and reduce false detection and missing detection. Finally,CutMix data is used to enhance and expand the forest fire smoke dataset,and the proposed method is evaluated. Theoretical analysis and experiments show that the improved model has significantly higher detection accuracy of forest fire smoke than that of the mainstream model. Compared with the traditional DETR model,the mAP (mean accuracy) of the model is increased by 4. 4% ,and the AP50 accuracy is increased by 3. 8%.

常用功能

更新日期/Last Update: 2025-02-10